Leadership 1 publisher2 min readPublished September 15, 2026 at 7:59 PM PDT

NIST is revising its AI risk management framework

NIST's own page says a revised AI RMF is in progress. The notice sits above the 2023 Core excerpt, and its four functions and subcategory numbering are what internal policies, audit files and vendor attestations quote.

The Board Room · Leadership desk

What happened

The 2023 Core divides govern, map, measure and manage into categories and subcategories, which are themselves subdivided into specific actions and outcomes.
The framework leaves users to decide which categories and subcategories they apply, and says its actions are not a checklist or an ordered set of steps.
NIST did not say when the revision will land or what will change, and has not published a draft of the revised framework.

Compiled by The Board RoomSomething wrong?How this is made

Why it matters

constraint If the revision renumbers categories or subcategories, every control matrix, audit workpaper and questionnaire that points at 1.0 identifiers has to be re-keyed. That work falls on whoever maintains the crosswalk.
decision Without a published date, waiting for the revised text is not a plan with an end. A program has to choose now between freezing its artifacts against 1.0 and keeping them loose enough to re-point later.
exposure An attestation that claims alignment to the NIST AI RMF without naming a version becomes ambiguous the day two versions exist. Resolving it falls to the customer holding the contract.
precedent Because the framework leaves selection and ordering to the adopter, the revision on its own sets no deadline; a deadline only arrives when a regulator or a buyer cites a particular version.

NIST's notice is two sentences: "The AI RMF 1.0 is being updated. A revised version is in progress." They sit above an excerpt labelled as coming from the 2023 framework, which is what the page still serves to anyone writing policy this week.

Governance paperwork rarely attaches to the four function names. It attaches a level down, where the Core breaks each function into categories and subcategories, and those into specific actions and outcomes. A policy that quotes govern, map, measure and manage survives a renumbering intact; a control matrix keyed to subcategory identifiers does not.

The 2023 text also limits how much a revision can strand. Some organizations select from among the categories and subcategories; others apply all of them, and NIST writes that the actions "do not constitute a checklist, nor are they necessarily an ordered set of steps". Once govern is instituted, most users would start with map and continue to measure or manage, though the functions may be performed in any order. The Playbook that supplies suggested tactical actions is voluntary as well, and both it and the framework sit in NIST's Trustworthy and Responsible AI Resource Center.

On this record, the revision of a voluntary framework is close to a documentation exercise. The cost appears where the version number was never written down. An attestation saying a vendor is aligned to the NIST AI RMF answers to one document today and to two after the revised version publishes. The party holding the contract is the one who has to ask which.

The govern function is where a rewrite would be felt first outside the compliance team. In 1.0 it is cross-cutting, infused throughout AI risk management, with compliance and evaluation aspects integrated into each of the other functions. It also addresses the full product lifecycle, including legal issues concerning third-party software or hardware systems and data. If the revision touches that language, the edit lands in supplier questionnaires.

Nothing expires this quarter, and the choice in front of a program is narrow: whether each artifact records the framework version and the subcategory it answers to. With both recorded, a version swap later is an edit to one column instead of a rereading of every policy.

What to watch

Publication of a draft or concept paper for the revised AI RMF, and whether the 1.0 subcategory numbering carries over into it.
Whether the Playbook is revised alongside the framework or trails it, leaving tactical actions keyed to the older text.
Whether customer contracts and supplier questionnaires start naming a specific AI RMF version and date instead of the framework generically.

Claim ledger

Ranked by verification strength, evidence, and original report placement.

[1]
A notice on NIST's AI RMF Core page states: "The AI RMF 1.0 is being updated. A revised version is in progress."
ReportedSupportedSource: airc.nist.gov, NIST AI RMF Core pageView cited source
[2]
The AI RMF Core is composed of four functions: govern, map, measure and manage. Each function is broken down into categories and subcategories, which are subdivided into specific actions and outcomes.
ReportedSupportedView cited source
[3]
The Core section on the page is presented as an excerpt from the NIST AI Risk Management Framework 1.0 (2023).
ReportedSupportedView cited source
[4]
The framework states that actions "do not constitute a checklist, nor are they necessarily an ordered set of steps".
ReportedSupportedView cited source
[5]
Framework users may apply the functions as best suits their needs based on resources and capabilities: some organizations may choose to select from among the categories and subcategories, others may apply all of them.
ReportedSupportedView cited source
[6]
Assuming a governance structure is in place, functions may be performed in any order across the AI lifecycle; after instituting the outcomes in govern, most users would start with the map function and continue to measure or manage.
ReportedSupportedView cited source
[7]
The NIST AI RMF Playbook is an online companion resource offering suggested tactical actions, and like the AI RMF it is voluntary.
ReportedSupportedView cited source
[8]
The Playbook and the AI RMF are both part of the NIST Trustworthy and Responsible AI Resource Center.
ReportedSupportedView cited source
[9]
Govern is described as a cross-cutting function infused throughout AI risk management that enables the other functions, and aspects of govern related to compliance or evaluation should be integrated into each of the other functions.
ReportedSupportedView cited source
[10]
The govern function addresses the full product lifecycle and associated processes, including legal and other issues concerning use of third-party software or hardware systems and data.
ReportedSupportedView cited source
[11]
The revision notice on the page appears without a date, a description of what will change, or a published draft.
ReportedSupportedView cited source
[12]
The govern function cultivates and implements a culture of risk management within organizations designing, developing, deploying, evaluating or acquiring AI systems.
ReportedSupportedView cited source
[13]
Conformance to AI RMF 1.0 is scoped by the adopting organization rather than fixed by the document, because category and subcategory selection is left to the user, the actions are not a checklist, and the companion Playbook is voluntary.
DerivedSupported1 source— sign in to open

Sources

Publishers with included, body-backed reporting in this cluster.

airc.nist.gov
1 article · September 15, 2026
AI RMF Core

Ask Clarity

Have a question this piece left open? The answer comes from its reporting and claim ledger — nothing else.

Try

Clarity analysis

How the coverage leans and what the record actually supports. Read the methodology

Perspective Coverage

How much of the coverage speaks to builders, operators and investors. It measures emphasis, not whether the story is right.

it arrives undated and without a draft

Reality

Evidence, adoption, hype gap, incentives and confidence, each read independently from the current record on the scale shown beneath it. How these are measured.

Evidence70

0100

One primary page, quoted directly

Every factual claim in this story is lifted from the document itself, including the revision notice word for word, so there is little room for transmission error about what version 1.0 says. The forward-looking half is thin by the same measure: one sentence, no draft, and no corroboration from anywhere outside NIST.

AdoptionInsufficient

No usage on the record

The page tells organizations how they may apply the functions, which is guidance and not a record of anyone applying them. To measure adoption we would need a usage disclosure or an attestation naming the framework, and this reporting includes neither.

Hype gap+15

−100 understated0 aligned+100 overstated

Framing runs slightly ahead of the page

On the framework's structure and on the notice itself, the reporting stays inside what NIST wrote. It goes further in describing internal policies, audit files and vendor attestations as quoting the four functions and the subcategory numbering: that is our inference about how the document circulates, and the page makes no such claim.

Incentives35

0100

Issuer describing its own voluntary framework

NIST wrote the framework, hosts the page and runs the resource center the Playbook sits in, so both the account of govern and the invitation to apply the functions selectively serve uptake of NIST's own document. No fee, license or product rides on it, which keeps the interest institutional: a voluntary framework matters only to the extent organizations pick it up.

Confidence60

0100

Firm on the text, weak on the change

What version 1.0 says can be checked against the document at any time, and this story quotes it. The claim readers will act on is the revision, and there the record is one line with nothing about scope or timing, from a single publisher.

Topics and entities

Follow any of these and your For You feed starts watching them — no settings page required.

Topics

Entities

invest
ClawHavoc's 12 accounts outshipped the malicious-skill census by 7.5 times
1 publisher
build
Fabricated SQLite CVEs cleared NVD, CISA ADP and Red Hat before anyone ran the code
1 publisher
product
An agent gets risky at the moment its recommendation becomes an outbound call
1 publisher
science
Three U.S. agencies name six China-based AI companies as distilling frontier models
1 publisher

What happened

Why it matters

What to watch

Claim ledger

Sources

Let Clarity write the post for you.

Topics and entities

Topics

Entities

Related stories

Claim ledger

Let Clarity write the post for you.

Perspective Coverage

Reality

Topics and entities

Topics

Entities

Related stories

Sources